Exploring Public Sentiment on Twitter: An NLP Approach¶

Part 1: Setting Up¶

In [1]:
# Importing InteractiveShell from IPython.core.interactiveshell module
from IPython.core.interactiveshell import InteractiveShell

# Setting the ast_node_interactivity option of InteractiveShell to "all"
# This allows IPython to display results for all statements in a code cell
# rather than just the last one, which is the default behavior.
InteractiveShell.ast_node_interactivity = "all"

import warnings
import matplotlib

# Suppress the specific MatplotlibDeprecationWarning
warnings.filterwarnings("ignore", category=matplotlib.cbook.MatplotlibDeprecationWarning)

/tmp/ipykernel_122/2357816857.py:13: MatplotlibDeprecationWarning: MatplotlibDeprecationWarning was deprecated in Matplotlib 3.6 and will be removed two minor releases later. Use matplotlib.MatplotlibDeprecationWarning instead.
  warnings.filterwarnings("ignore", category=matplotlib.cbook.MatplotlibDeprecationWarning)

Task 1: Import the Libraries¶

In [2]:
import os  
import re 
import pandas as pd  
import numpy as np 
import seaborn as sns  
import matplotlib.pyplot as plt 
from nltk.tokenize import word_tokenize
from nltk.corpus import stopwords
from nltk.stem import WordNetLemmatizer, PorterStemmer
from nltk.sentiment.vader import SentimentIntensityAnalyzer
from sklearn.decomposition import LatentDirichletAllocation
from sklearn.feature_extraction.text import CountVectorizer
import plotly_express as px
from gensim.corpora import Dictionary
from gensim.models.ldamulticore import LdaMulticore
from gensim.models.coherencemodel import CoherenceModel
import pyLDAvis.gensim
from wordcloud import WordCloud
import nbconvert

Part 2: Data Collection and Preprocessing¶

Task 2: Load the Dataset and Have a First Look¶

Load the CSV into a variable named df_twitter¶
In [3]:
df_twitter = pd.read_csv('./usercode/covid19_twitter_dataset.csv', index_col=0)
Display the DataFrame along with the row count¶
In [4]:
df_twitter.head()
df_twitter.shape
Out[4]:
user_name user_location user_description user_created user_followers user_friends user_favourites user_verified date text hashtags source is_retweet language lat long country
0 Tom Basile 🇺🇸 new york, ny Husband, Father, Columnist & Commentator. Auth... 2009-04-16 20:06:23 2253 1677 24 True 2020-07-25 12:27:17 Hey @Yankees @YankeesPR and @MLB - wouldn't it... NaN Twitter for Android False en 40.712728 -74.006015 United States
1 Time4fisticuffs pewee valley, ky #Christian #Catholic #Conservative #Reagan #Re... 2009-02-28 18:57:41 9275 9525 7254 False 2020-07-25 12:27:14 @diane3443 @wdunlap @realDonaldTrump Trump nev... ['COVID19'] Twitter for Android False en 38.310625 -85.487459 United States
2 DIPR-J&K jammu and kashmir 🖊️Official Twitter handle of Department of Inf... 2017-02-12 06:45:15 101009 168 101 False 2020-07-25 12:27:08 25 July : Media Bulletin on Novel #CoronaVirus... ['CoronaVirusUpdates', 'COVID19'] Twitter for Android False en 33.664930 75.162958 India
3 🎹 Franz Schubert новоро́ссия 🎼 #Новоро́ссия #Novorossiya #оставайсядома #S... 2018-03-19 16:29:52 1180 1071 1287 False 2020-07-25 12:27:06 #coronavirus #covid19 deaths continue to rise.... ['coronavirus', 'covid19'] Twitter Web App False en 43.341088 132.625674 Россия
5 Creativegms dhaka,bangladesh I'm Motalib Mia, Logo -Logo Designer - Brandin... 2020-01-12 09:03:01 241 1694 8443 False 2020-07-25 12:26:50 Order here: https://t.co/4NUrGX6EmA\n\n#logo #... ['logo', 'graphicdesigner', 'logodesign', 'log... Twitter Web App False en 23.764402 90.389015 বাংলাদেশ
Out[4]:
(111973, 17)
In [5]:
df_twitter.columns
Out[5]:
Index(['user_name', 'user_location', 'user_description', 'user_created',
       'user_followers', 'user_friends', 'user_favourites', 'user_verified',
       'date', 'text', 'hashtags', 'source', 'is_retweet', 'language', 'lat',
       'long', 'country'],
      dtype='object')
Show the tweet count for the top 10 countries¶
In [6]:
df_twitter['country'].value_counts()[:10]
Out[6]:
United States     41931
India             19473
United Kingdom    11544
Canada             6679
Australia          4370
Nigeria            2632
South Africa       2415
Éire / Ireland     1545
Kenya              1493
中国                 1141
Name: country, dtype: int64
Plot the top 20 users who post the most¶
In [7]:
df_twitter['user_name'].value_counts()[:20].plot(kind='barh')
Out[7]:
<Axes: >
No description has been provided for this image

Task 3: Basic Text Preprocessing¶

Remove unnecessary columns for the analysis¶
In [8]:
df_twitter.drop(['user_description','user_created','user_favourites','language'], axis=1, inplace=True)
Check for missing values and handle them¶
In [9]:
df_twitter.isnull().sum() # to check for missing values

# Fill missing values in 'hashtags' with an empty list
df_twitter['hashtags'].fillna("[]", inplace=True)
Out[9]:
user_name             0
user_location         0
user_followers        0
user_friends          0
user_verified         0
date                  0
text                  0
hashtags          32184
source                0
is_retweet            0
lat                   0
long                  0
country               0
dtype: int64
Convert date column to datetime object and extract features¶
In [10]:
df_twitter['date'] = pd.to_datetime(df_twitter['date'])
df_twitter['year'] = df_twitter['date'].dt.year
df_twitter['month'] = df_twitter['date'].dt.month
df_twitter['day'] = df_twitter['date'].dt.day
df_twitter['hour'] = df_twitter['date'].dt.hour
df_twitter['day_of_week'] = df_twitter['date'].dt.dayofweek
In [11]:
df_twitter.head()
Out[11]:
user_name user_location user_followers user_friends user_verified date text hashtags source is_retweet lat long country year month day hour day_of_week
0 Tom Basile 🇺🇸 new york, ny 2253 1677 True 2020-07-25 12:27:17 Hey @Yankees @YankeesPR and @MLB - wouldn't it... [] Twitter for Android False 40.712728 -74.006015 United States 2020 7 25 12 5
1 Time4fisticuffs pewee valley, ky 9275 9525 False 2020-07-25 12:27:14 @diane3443 @wdunlap @realDonaldTrump Trump nev... ['COVID19'] Twitter for Android False 38.310625 -85.487459 United States 2020 7 25 12 5
2 DIPR-J&K jammu and kashmir 101009 168 False 2020-07-25 12:27:08 25 July : Media Bulletin on Novel #CoronaVirus... ['CoronaVirusUpdates', 'COVID19'] Twitter for Android False 33.664930 75.162958 India 2020 7 25 12 5
3 🎹 Franz Schubert новоро́ссия 1180 1071 False 2020-07-25 12:27:06 #coronavirus #covid19 deaths continue to rise.... ['coronavirus', 'covid19'] Twitter Web App False 43.341088 132.625674 Россия 2020 7 25 12 5
5 Creativegms dhaka,bangladesh 241 1694 False 2020-07-25 12:26:50 Order here: https://t.co/4NUrGX6EmA\n\n#logo #... ['logo', 'graphicdesigner', 'logodesign', 'log... Twitter Web App False 23.764402 90.389015 বাংলাদেশ 2020 7 25 12 5
Apply basic_clean_text() function to text column¶
In [12]:
def basic_clean_text(text):
    # Convert to lowercase
    text = text.lower()
    
    # Remove extra whitespace
    text = re.sub(r" +", ' ', text, flags=re.MULTILINE)
    
    # Remove URLs
    text = re.sub(r"http\S+|www\S+|https\S+", '', text, flags=re.MULTILINE)
    
    # remove HTML tags
    text = text.replace('<.*?>', '')  
    
     # remove non-alphanumeric characters
    text = text.replace('[^A-Za-z0-9 ]+', '') 
    
    # Remove punctuation
    text = re.sub(r'[^\w\s]', '', text)
    
    # Remove numbers
    text = re.sub(r'\d+', '', text)
    
    return text

df_twitter['text'] = df_twitter['text'].apply(basic_clean_text)
In [13]:
df_twitter['text'].head()
Out[13]:
0    hey yankees yankeespr and mlb  wouldnt it have...
1    diane wdunlap realdonaldtrump trump never once...
2     july  media bulletin on novel coronavirusupda...
3    coronavirus covid deaths continue to rise its ...
5    order here \n\nlogo graphicdesigner logodesign...
Name: text, dtype: object

Task 4: Implement Advanced Text Preprocessing¶

Apply advanced_text_preprocessing() function to text column¶
In [14]:
def advanced_text_preprocessing(text):
    # Tokenization
    tokens = word_tokenize(text)

    # Stopword Removal
    stop_words = set(stopwords.words('english'))
    filtered_tokens = [word for word in tokens if word.lower() not in stop_words]

    # Lemmatization
    lemmatizer = WordNetLemmatizer()
    lemmatized_tokens = [lemmatizer.lemmatize(word) for word in filtered_tokens]

    # Join tokens back to form the preprocessed text
    preprocessed_text = " ".join(lemmatized_tokens)

    return preprocessed_text

df_twitter['text'] = df_twitter['text'].apply(advanced_text_preprocessing)
In [15]:
df_twitter['text'].head()
Out[15]:
0    hey yankee yankeespr mlb wouldnt made sense pl...
1    diane wdunlap realdonaldtrump trump never clai...
2    july medium bulletin novel coronavirusupdates ...
3    coronavirus covid death continue rise almost b...
5    order logo graphicdesigner logodesign logodesi...
Name: text, dtype: object

Part 3: Sentiment Analysis¶

Task 5: Perform Sentiment Analysis with Vader Library¶

Perform sentiment analysis with vader library¶
In [16]:
# Initialize VADER
sid = SentimentIntensityAnalyzer()

# Get sentiment scores for each tweet
df_twitter['sentiment_scores'] = df_twitter['text'].apply(lambda x: sid.polarity_scores(x))
Display a random sample of 10 tweets with their sentiment scores¶
In [17]:
df_twitter[['text','sentiment_scores']].sample(10).values
Out[17]:
array([['total number covid test conducted far lagos stand confirmed case',
        {'neg': 0.0, 'neu': 0.874, 'pos': 0.126, 'compound': 0.0772}],
       ['mron rahulgandhi done parliamentary constituency help citizen covid',
        {'neg': 0.0, 'neu': 0.722, 'pos': 0.278, 'compound': 0.4019}],
       ['madurai crossing covid case death rate highest tamilnadu higher chennai well',
        {'neg': 0.26, 'neu': 0.6, 'pos': 0.14, 'compound': -0.4215}],
       ['covid ucsandiego chronicle august ucsdhealth number continue improve week',
        {'neg': 0.0, 'neu': 0.625, 'pos': 0.375, 'compound': 0.4939}],
       ['dpharmaplc serestx well done involved waiting eagerly news whether microbiome worki',
        {'neg': 0.0, 'neu': 0.657, 'pos': 0.343, 'compound': 0.5719}],
       ['beautiful nurse heart stopped beating covid striked mirian awour adumba put go',
        {'neg': 0.275, 'neu': 0.506, 'pos': 0.219, 'compound': 0.0}],
       ['today covid update ohio live today tie hiramcollege',
        {'neg': 0.0, 'neu': 1.0, 'pos': 0.0, 'compound': 0.0}],
       ['hope today watchman going arrested stamped gate pas kemsa covidmillionaires enoughisenough covid',
        {'neg': 0.194, 'neu': 0.625, 'pos': 0.181, 'compound': -0.0516}],
       ['madagascar president andry rajoelina say nation win covid launch covdata digital mgt',
        {'neg': 0.0, 'neu': 0.743, 'pos': 0.257, 'compound': 0.5859}],
       ['covid total covid case recovery total death last hour ca',
        {'neg': 0.302, 'neu': 0.698, 'pos': 0.0, 'compound': -0.5994}]],
      dtype=object)

Task 6: Classify the Tweets into Positive, Neutral and Negative¶

Classify the tweets into categories of positive, negative, or neutral sentiment¶
In [18]:
threshold_value = 0.0

# Get sentiment based on the compound score
df_twitter['sentiment'] = df_twitter['sentiment_scores'].apply(lambda x: 'positive' if x['compound'] > threshold_value else ('neutral' if x['compound'] == threshold_value else 'negative'))

# Print the sentiment distribution
print(df_twitter['sentiment'].value_counts())

positive    45225
negative    33807
neutral     32941
Name: sentiment, dtype: int64

Part 4: Trend Analysis and Visualization¶

Task 7: Display the Evolution of Sentiment Over Time¶

Extract the top 3 countries to be used as a filter¶
In [19]:
# Identifying the 3 most active countries based on the number of tweets
most_active_countries = df_twitter['country'].value_counts().nlargest(3).index.tolist()
most_active_countries
Out[19]:
['United States', 'India', 'United Kingdom']
Filter the DataFrame¶
In [20]:
# Filtering the data to include only the 3 most active countries
filtered_data = df_twitter[df_twitter['country'].isin(most_active_countries)]
filtered_data.shape
Out[20]:
(72948, 20)
Create sentiment over time by country based on groupby¶
In [21]:
sentiment_over_time_by_country = filtered_data.groupby([pd.Grouper(key='date', freq='D'), 'country'])['sentiment'].value_counts().unstack().fillna(0).reset_index()

# Melt the DataFrame to have a 'sentiment' column and a 'count' column
sentiment_melted = sentiment_over_time_by_country.melt(id_vars=['date', 'country'], value_vars=['negative', 'neutral', 'positive'], var_name='sentiment', value_name='count')
In [22]:
sentiment_melted.head()
Out[22]:
date country sentiment count
0 2020-07-24 India negative 1.0
1 2020-07-24 United Kingdom negative 4.0
2 2020-07-24 United States negative 49.0
3 2020-07-25 India negative 449.0
4 2020-07-25 United Kingdom negative 328.0
Plot the top three countries sentiments over time¶
In [23]:
# For each country, create a separate plot
for country in most_active_countries:
    _ = plt.figure(figsize=(15, 6))
    _ = sns.lineplot(data=sentiment_melted[sentiment_melted['country'] == country], x="date", y="count", hue="sentiment")
    _ = plt.title(f'Sentiment Counts Over Time for {country}')
    _ = plt.show()
No description has been provided for this image
No description has been provided for this image
No description has been provided for this image

Task 8: Use Wordcloud to Visualize Words Used in Sentiments¶

Define the create_word_cloud() function¶
In [24]:
# Function to create word cloud for a given sentiment
def create_word_cloud(sentiment):
    # Filtering the text based on the given sentiment
    text = ' '.join(df_twitter[df_twitter['sentiment'] == sentiment]['text'].values)
    
    # Creating the word cloud
    wordcloud = WordCloud(width=800, height=400, background_color='white').generate(text)
    
    # Plotting the word cloud
    plt.figure(figsize=(15, 6))
    plt.imshow(wordcloud, interpolation='bilinear')
    plt.axis("off")
    plt.title(f'Word Cloud for {sentiment.capitalize()} Sentiments')
    plt.show()
Creating word cloud for positive sentiment¶
In [25]:
create_word_cloud('positive')

/usr/local/lib/python3.8/dist-packages/wordcloud/wordcloud.py:106: MatplotlibDeprecationWarning: The get_cmap function was deprecated in Matplotlib 3.7 and will be removed two minor releases later. Use ``matplotlib.colormaps[name]`` or ``matplotlib.colormaps.get_cmap(obj)`` instead.
  self.colormap = plt.cm.get_cmap(colormap)
No description has been provided for this image
Creating word cloud for negative sentiment¶
In [26]:
create_word_cloud('negative')

/usr/local/lib/python3.8/dist-packages/wordcloud/wordcloud.py:106: MatplotlibDeprecationWarning: The get_cmap function was deprecated in Matplotlib 3.7 and will be removed two minor releases later. Use ``matplotlib.colormaps[name]`` or ``matplotlib.colormaps.get_cmap(obj)`` instead.
  self.colormap = plt.cm.get_cmap(colormap)
No description has been provided for this image
Creating word cloud for neutral sentiment¶
In [27]:
create_word_cloud('neutral')

/usr/local/lib/python3.8/dist-packages/wordcloud/wordcloud.py:106: MatplotlibDeprecationWarning: The get_cmap function was deprecated in Matplotlib 3.7 and will be removed two minor releases later. Use ``matplotlib.colormaps[name]`` or ``matplotlib.colormaps.get_cmap(obj)`` instead.
  self.colormap = plt.cm.get_cmap(colormap)
No description has been provided for this image

Task 9: Display the Sentiment on a Geographical Heatmap¶

Create mapping from positive, neutral and negative to an numerical value for visualization purpose¶
In [28]:
sentiment_mapping = {'positive': 1, 'neutral': 0, 'negative': -1}
df_twitter['sentiment_value'] = df_twitter['sentiment'].map(sentiment_mapping)
In [29]:
df_twitter.head()
Out[29]:
user_name user_location user_followers user_friends user_verified date text hashtags source is_retweet ... long country year month day hour day_of_week sentiment_scores sentiment sentiment_value
0 Tom Basile 🇺🇸 new york, ny 2253 1677 True 2020-07-25 12:27:17 hey yankee yankeespr mlb wouldnt made sense pl... [] Twitter for Android False ... -74.006015 United States 2020 7 25 12 5 {'neg': 0.112, 'neu': 0.64, 'pos': 0.248, 'com... positive 1
1 Time4fisticuffs pewee valley, ky 9275 9525 False 2020-07-25 12:27:14 diane wdunlap realdonaldtrump trump never clai... ['COVID19'] Twitter for Android False ... -85.487459 United States 2020 7 25 12 5 {'neg': 0.0, 'neu': 0.832, 'pos': 0.168, 'comp... positive 1
2 DIPR-J&K jammu and kashmir 101009 168 False 2020-07-25 12:27:08 july medium bulletin novel coronavirusupdates ... ['CoronaVirusUpdates', 'COVID19'] Twitter for Android False ... 75.162958 India 2020 7 25 12 5 {'neg': 0.0, 'neu': 0.796, 'pos': 0.204, 'comp... positive 1
3 🎹 Franz Schubert новоро́ссия 1180 1071 False 2020-07-25 12:27:06 coronavirus covid death continue rise almost b... ['coronavirus', 'covid19'] Twitter Web App False ... 132.625674 Россия 2020 7 25 12 5 {'neg': 0.433, 'neu': 0.488, 'pos': 0.079, 'co... negative -1
5 Creativegms dhaka,bangladesh 241 1694 False 2020-07-25 12:26:50 order logo graphicdesigner logodesign logodesi... ['logo', 'graphicdesigner', 'logodesign', 'log... Twitter Web App False ... 90.389015 বাংলাদেশ 2020 7 25 12 5 {'neg': 0.0, 'neu': 1.0, 'pos': 0.0, 'compound... neutral 0

5 rows × 21 columns

Visualize a geographical heatmap of all sentiment on the map¶
In [30]:
fig = px.density_mapbox(df_twitter, lat='lat', lon='long',
                        z='sentiment_value', radius=20,
                        center=dict(lat=df_twitter.lat.mean(), 
                                    lon=df_twitter.long.mean()), 
                        zoom=4,
                        mapbox_style="open-street-map", 
                        height=900)
fig.show()

Part 5: Topic Modeling¶

Task 10: Train the LDA (Gensim) Model¶

Preprocess the text¶
In [31]:
df_twitter['text_tokens'] = df_twitter['text'].str.lower().str.split()
Create a dictionary¶
In [32]:
id2word = Dictionary(df_twitter['text_tokens'])
Filter extremes¶
In [33]:
id2word.filter_extremes(no_below=2, no_above=.99)
Create a corpus¶
In [34]:
corpus = [id2word.doc2bow(d) for d in df_twitter['text_tokens']]
Instantiate an LDA model¶
In [35]:
base_model = LdaMulticore(corpus=corpus, num_topics=5, id2word=id2word, workers=12, 
passes=5)

Task 11: Evaluate the Model¶

Print the topics¶
In [36]:
# Filtering for words 
words = [re.findall(r'"([^"]*)"',t[1]) for t in base_model.print_topics()]

# Create Topics
topics = [' '.join(t[0:10]) for t in words]

# Getting the topics
for id, t in enumerate(topics): 
    print(f"------ Topic {id} ------")
    print(t, end="\n\n")

------ Topic 0 ------
covid new case coronavirus death today amp people health first

------ Topic 1 ------
covid mask amp people u pandemic school coronavirus vaccine time

------ Topic 2 ------
covid amp one help trump realdonaldtrump people even know risk

------ Topic 3 ------
covid people pandemic like dont take get health u care

------ Topic 4 ------
covid case death india day last coronavirus new positive total
Compute the perplexity and coherence score¶
In [37]:
# Compute Perplexity
# a measure of how good the model is. lower the better
base_perplexity = base_model.log_perplexity(corpus)
print('\nPerplexity: ', base_perplexity) 

# Compute Coherence Score
coherence_model = CoherenceModel(model=base_model, texts=df_twitter['text_tokens'], 
                                   dictionary=id2word, coherence='c_v')
coherence_lda_model_base = coherence_model.get_coherence()
print('\nCoherence Score: ', coherence_lda_model_base)

Perplexity:  -8.413091135076575

Coherence Score:  0.3449902195784292

Task 12: Classify Twitter Tweets into Topics¶

Topic classification function¶
In [38]:
def classify_tweet(tweet):
    # Preprocess the tweet
    processed_tweet = tweet.lower().split()

    # Transform tweet into bag-of-words format
    tweet_bow = id2word.doc2bow(processed_tweet)

    # Get topic probabilities
    topic_probabilities = base_model.get_document_topics(tweet_bow)

    # Get the topic with the highest probability
    most_likely_topic = max(topic_probabilities, key=lambda x: x[1])

    return most_likely_topic[0]
Classify all tweets¶
In [39]:
df_twitter['topic'] = df_twitter['text'].apply(lambda x: classify_tweet(x))
In [40]:
df_twitter['topic'].value_counts()
Out[40]:
1    26131
2    24087
4    21727
3    21562
0    18466
Name: topic, dtype: int64
Examine topic distribution¶
In [ ]:

Part 6: Interpretation of Results¶

Task 13: Identify Relationships between Sentiment and Topic¶

Grouping and aggregating data by topic and sentiment¶
In [41]:
grouped = df_twitter.groupby(['topic', 'sentiment']).size().unstack(level='sentiment')
print(grouped)

sentiment  negative  neutral  positive
topic                                 
0              5188     6813      6465
1              5369     7941     12821
2              7930     5916     10241
3              7379     5084      9099
4              7941     7187      6599
Calculating proportions¶
In [42]:
percent_grouped = grouped.divide(grouped.sum(axis=1), axis=0)
print(percent_grouped)

sentiment  negative   neutral  positive
topic                                  
0          0.280949  0.368948  0.350103
1          0.205465  0.303892  0.490643
2          0.329223  0.245610  0.425167
3          0.342222  0.235785  0.421992
4          0.365490  0.330787  0.303723
Visualizing the results¶
In [43]:
# Assign the colors based on your data's columns (assuming the columns of percent_grouped are 'positive', 'neutral', and 'negative')
palette = {'positive': '#66BB6A', 'neutral': '#BDBDBD', 'negative': '#EF5350'}
colors = [palette[col] for col in percent_grouped.columns]

percent_grouped.plot(kind='bar', stacked=True, color=colors)
plt.xlabel('Topic')
plt.ylabel('Proportion of Tweets')
plt.title('Proportion of Sentiments by Topic')
plt.legend(loc='upper right')
plt.show()
Out[43]:
<Axes: xlabel='topic'>
Out[43]:
Text(0.5, 0, 'Topic')
Out[43]:
Text(0, 0.5, 'Proportion of Tweets')
Out[43]:
Text(0.5, 1.0, 'Proportion of Sentiments by Topic')
Out[43]:
<matplotlib.legend.Legend at 0x72c031fc0640>
No description has been provided for this image

Task 14: Interpret the Topic Modeling Results¶

Creating topic distance visualization¶
In [44]:
pyLDAvis.enable_notebook()
pyLDAvis.gensim.prepare(base_model, corpus, id2word)

/usr/local/lib/python3.8/dist-packages/pyLDAvis/_prepare.py:246: FutureWarning:

In a future version of pandas all arguments of DataFrame.drop except for the argument 'labels' will be keyword-only.
Out[44]:

Task 15: Compile your Findings into a Final Report with NBConvert¶

Execute the command within the provided notebook cell¶
In [ ]:
!jupyter nbconvert --to html ../usercode/CovidSentimentAnalyzer.ipynb

CONGRATULATIONS¶